Dear Challenge Organisers, Regarding the validation data, the [Challenge Questions and Scoring](https://www.synapse.org/#!Synapse:syn15589870/wiki/592683) section states that **"Immune and stromal cells profiled for validation data were derived from healthy donors"**, whereas the [Input files](https://www.synapse.org/#!Synapse:syn15589870/wiki/592699) section shows that the **cancer.type** for validation data will be **"BRCA or CRC"**. Are these statements both within the context of data for the validation phase? If so, can you please explain the discrepancy between the two? In addition, since the models are supposed to also predict fibroblasts and endothelial cells, can we assume that all leaderboard and validation data came from tissue samples and not PBMCs? This would be very helpful for deriving the correct gene expression profiles. Thank you.

Created by Miro Phan miro-beda
Hi Brian, Thank you! That clears things up. Best, Miro
Hi @miro-beda , Yes, both statements are in the context of the validation data. The admixtures include immune cells, fibroblasts and endothelial cells, and cancer/malignant cells. The immune cells, fibroblasts, and endothelial cells are from healthy donors (not necessarily the same donor--i.e., cell type A may be from donor X and cell type B may be from donor Y). You should _not_ assume that all cells are from tissue samples. Though you are correct that the fibroblasts and endothelial cells were either directly or indirectly (i.e., cell line-based) derived from tissue. When we indicate that cancer.type = BRCA, we mean only that the cancer cells represent breast cancer (i.e., a breast cancer patient or a breast cancer cell line) and _not_ that the immune, fibroblast, or endothelial cells are derived from a breast cancer patient or cell line. We recognize that there are several unfortunately unrealistic things going on here -- cells from multiple patients / sources being mixed together and cells from healthy and cancer patients mixed together. This approach does allow for a definitive ground truth. We will address more realistic admixtures (i.e., directly derived from tissue) in the subsequent "in vivo challenge" (which we have yet to announce). Thanks for your questions and I'm sorry for the ambiguity. Let me know if we haven't clarified. Best, Brian

Validation data derived from healthy donors or cancer patients? page is loading…